Deep multi-omics integration by learning correlation-maximizing representation identifies prognostically stratified cancer subtypes.
Deep multi-omics integration by learning correlation-maximizing representation identifies prognostically stratified cancer subtypes.
复制标题
DOI：
10.1093/bioadv/vbad075
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
Davuluri, Ramana
中科院分区：
文献类型：
--
作者：
Ji, Yanrong;Dutta, Pratik;Davuluri, Ramana
通过多组学和临床数据的综合建模进行分子亚型分型可以帮助识别稳健且临床上可行的疾病亚组；开发精准医学方法的重要一步。我们开发了一种新颖的以结果为导向的分子亚组框架，称为通过最大化相关性进行深度多组学综合亚型分析（DeepMOIS-MC），通过最大化所有输入组学视图之间的相关性来从多组学数据中进行综合学习。 DeepMOIS-MC由两部分组成：聚类和分类。在聚类部分，将预处理后的高维多组学视图输入到两层全连接神经网络中。各个网络的输出受到广义典型相关分析损失的影响，以学习共享表示。接下来，通过回归模型对学习到的表示进行过滤，以选择与协变量临床变量相关的特征，例如生存/结果。过滤后的特征用于聚类以确定最佳聚类分配。在分类阶段，基于等频分箱对组学视图之一的原始特征矩阵进行缩放和离散化，然后使用随机森林进行特征选择。使用这些选定的特征，构建分类模型（例如 XGBoost 模型）来预测在聚类阶段识别的分子亚组。我们使用 TCGA 数据集将 DeepMOIS-MC 应用于肺癌和肝癌。在比较分析中，我们发现 DeepMOIS-MC 在患者分层方面优于传统方法。最后，我们在独立数据集上验证了分类模型的稳健性和通用性。我们预计 DeepMOIS-MC 可以用于许多多组学综合分析任务。 DGCCA 和其他 DeepMOIS-MC 模块的 PyTorch 实现的源代码可在 GitHub (https://github.com/duttaprat/DeepMOIS-MC) 上找到。 补充数据可在生物信息学进展在线获取。
Molecular subtyping by integrative modeling of multi-omics and clinical data can help the identification of robust and clinically actionable disease subgroups; an essential step in developing precision medicine approaches. We developed a novel outcome-guided molecular subgrouping framework, called Deep Multi-Omics Integrative Subtyping by Maximizing Correlation (DeepMOIS-MC), for integrative learning from multi-omics data by maximizing correlation between all input -omics views. DeepMOIS-MC consists of two parts: clustering and classification. In the clustering part, the preprocessed high-dimensional multi-omics views are input into two-layer fully connected neural networks. The outputs of individual networks are subjected to Generalized Canonical Correlation Analysis loss to learn the shared representation. Next, the learned representation is filtered by a regression model to select features that are related to a covariate clinical variable, for example, a survival/outcome. The filtered features are used for clustering to determine the optimal cluster assignments. In the classification stage, the original feature matrix of one of the -omics view is scaled and discretized based on equal frequency binning, and then subjected to feature selection using RandomForest. Using these selected features, classification models (for example, XGBoost model) are built to predict the molecular subgroups that were identified at clustering stage. We applied DeepMOIS-MC on lung and liver cancers, using TCGA datasets. In comparative analysis, we found that DeepMOIS-MC outperformed traditional approaches in patient stratification. Finally, we validated the robustness and generalizability of the classification models on independent datasets. We anticipate that the DeepMOIS-MC can be adopted to many multi-omics integrative analyses tasks. Source codes for PyTorch implementation of DGCCA and other DeepMOIS-MC modules are available at GitHub (https://github.com/duttaprat/DeepMOIS-MC). Supplementary data are available at Bioinformatics Advances online.
登录
查看更多内容
鸟类多样性的密集采样增强了比较基因组学的力量
影响因子：
64.8
作者：
Feng S;Stiller J;Deng Y;Armstrong J;Fang Q;Reeve AH;Xie D;Chen G;Guo C;Faircloth BC;Petersen B;Wang Z;Zhou Q;Diekhans M;Chen W;Andreu-Sánchez S;Margaryan A;Howard JT;Parent C;Pacheco G;Sinding MS;Puetz L;Cavill E;Ribeiro ÂM;Eckhart L;Fjeldså J;Hosner PA;Brumfield RT;Christidis L;Bertelsen MF;Sicheritz-Ponten T;Tietze DT;Robertson BC;Song G;Borgia G;Claramunt S;Lovette IJ;Cowen SJ;Njoroge P;Dumbacher JP;Ryder OA;Fuchs J;Bunce M;Burt DW;Cracraft J;Meng G;Hackett SJ;Ryan PG;Jønsson KA;Jamieson IG;da Fonseca RR;Braun EL;Houde P;Mirarab S;Suh A;Hansson B;Ponnikas S;Sigeman H;Stervander M;Frandsen PB;van der Zwan H;van der Sluis R;Visser C;Balakrishnan CN;Clark AG;Fitzpatrick JW;Bowman R;Chen N;Cloutier A;Sackton TB;Edwards SV;Foote DJ;Shakya SB;Sheldon FH;Vignal A;Soares AER;Shapiro B;González-Solís J;Ferrer-Obiol J;Rozas J;Riutort M;Tigano A;Friesen V;Dalén L;Urrutia AO;Székely T;Liu Y;Campana MG;Corvelo A;Fleischer RC;Rutherford KM;Gemmell NJ;Dussex N;Mouritsen H;Thiele N;Delmore K;Liedvogel M;Franke A;Hoeppner MP;Krone O;Fudickar AM;Milá B;Ketterson ED;Fidler AE;Friis G;Parody-Merino ÁM;Battley PF;Cox MP;Lima NCB;Prosdocimi F;Parchman TL;Schlinger BA;Loiselle BA;Blake JG;Lim HC;Day LB;Fuxjager MJ;Baldwin MW;Braun MJ;Wirthlin M;Dikow RB;Ryder TB;Camenisch G;Keller LF;DaCosta JM;Hauber ME;Louder MIM;Witt CC;McGuire JA;Mudge J;Megna LC;Carling MD;Wang B;Taylor SA;Del-Rio G;Aleixo A;Vasconcelos ATR;Mello CV;Weir JT;Haussler D;Li Q;Yang H;Wang J;Lei F;Rahbek C;Gilbert MTP;Graves GR;Jarvis ED;Paten B;Zhang G
通讯作者：
Zhang G
Biopython：用于计算分子生物学和生物信息学的免费 Python 工具。
DOI：
10.1093/bioinformatics/btp163
发表时间：
2009-06-01
期刊：
Bioinformatics (Oxford, England)
影响因子：
--
作者：
Cock PJ;Antao T;Chang JT;Chapman BA;Cox CJ;Dalke A;Friedberg I;Hamelryck T;Kauff F;Wilczynski B;de Hoon MJ
通讯作者：
de Hoon MJ
影响因子：
1.3
作者：
Pal, Subhajit;Mondal, Sudip;Ghosh, Zhumur
通讯作者：
Ghosh, Zhumur
影响因子：
16.8
作者：
Jetz, Walter;Pyron, R. Alexander
通讯作者：
Pyron, R. Alexander
影响因子：
3
作者：
Boyle B;Hopkins N;Lu Z;Raygoza Garay JA;Mozzherin D;Rees T;Matasci N;Narro ML;Piel WH;McKay SJ;Lowry S;Freeland C;Peet RK;Enquist BJ
通讯作者：
Enquist BJ